“Trust is earned. I’m not going to rely on the cloud to guess something mission critical on our ‘first date’, just as I’m unlikely to do a lot of business with someone I’ve just met. I want to know it knows me – to feel understood – before I start betting on it.” #mexuservoice

At the intersection of artificial intelligence and user experience is a tension: to what degree does the designer – and by extension, the user – trust a machine to make a good decision?

The complexities of artificial intelligence, wondrous as they are, can sometimes obscure the practicalities, so let’s explore a straightforward example: how much of a user interface should remain static and how much should be dynamic, controlled by the decisions of an artificial intelligence engine?

The technology already exists today to change the user interface of digital experiences in real time, responding directly to user behaviour. However, to do so carries inherent risks: if the machine guesses incorrectly, the user experience suffers and trust is reduced. Muscle memory, whereby users establish reliable patterns which allow them to move at speed through familiar interfaces, is also impacted.

Our MEX explorations have led us to this principle, whereby the use of artificial intelligence is governed by an etiquette similar to human relationships. It relies upon the gradual building of trust through numerous small interactions and does not seek a single, transformational moment to win over the user.

For further views on this theme, try:

• Episode 10 of the MEX podcast, where we talk to Nathan Benaich, partner at Playfair Capital, a VC specialising in AI investments
• How artificial intelligence nuances experience design, part of the MEX Friday inspirations series
• Episode 2 of the MEX podcast, talking to Ed Rex, CEO of Jukedeck, an artificial intelligence engine for music
• MEX Pathway #14, cataloguing ongoing work into the principles for context aware design

The principle, part of an emerging series in the MEX journal, is summarised below in a tweetable, shareable graphic. Please share and thank you for citing appropriately.